The Analysis of Verbal Behavior 1987, 5, 33-39

Rule-Following and Human Operant Responding:

Conceptual and Methodological Considerations
Robert D. Zettle and Mark J. Young
Wichita State University
A conceptual analysis of rule-govemed behavior, emphasizing pliance and tracking as functional
classes of rule-following, is provided and related to previous methodological strategies in human
operant research. A novel strategy, which utilizes a microcomputer to reinforce correspondences
between subject guesses and responding, is proposed for the study of rule-following. Results
from a preliminary demonstration of the procedures are reported briefly, and possible applica-
tions to the further analysis of rule-following are discussed.

Responding of human subjects on operant
schedules presumably is multiply deter-
mined. In addition to being shaped directly
by programmed contingencies, human oper-
ant responding may constitute instances of
rule-governed behavior. The last decade in
particular has seen increased attention given
to the potential role of verbal control in
human operant behavior (e.g., Baron &
Galizio, 1983; Harzem, Lowe, & Bagshaw,
1978; Lowe, 1979, 1983). Whether or not a
particular example of human operant
responding is contingency shaped and/or
rule-governed cannot be determined a priori
but must be based on a functional analysis
of the behavior in question.
Because of the fine-grained analysis of
behavior which operant schedules permit,
they provide an especially useful preparation
for the study of rule following. Several
strategies have been employed in this under-
taking. These have included the systematic
presentation and withdrawal of instructions
which may or may not correspond to pro-
grammed contingencies (e.g., Galizio, 1979,
Hayes, Brownstein, Zettle, Rosenfarb, Korn,
1986), shaping or instructing subjects' verbal
behavior (Catania, Matthews, & Shimoff,
1982; Matthews, Catania, & Shimoff, 1985),
the analysis of postsession verbal reports
(Shimoff, 1986), and performance compari-
sons of preverbal and verbal children on
operant schedules (Bentall, Lowe, & Beasty,
Reprint requests should be sent to Robert D. Zettle,
Psychology Department, Wichita State University,
Wichita, Kansas 67208.
33
1985; Lowe, Beasty, & Bentall, 1983). The
purpose of this paper is to review earlier
work in the context of a specific concep-
tualization of rule-governed behavior. An
additional strategy for the analysis of verbal
control in human operant responding will be
described and results from a preliminary
study will be discussed.
FUNCTIONAL CLASSES
OF RULE-FOLLOWING
For present purposes at least two distinct
functional classes of rule-governed behavior,
pliance and tracking, may be considered
(Zettle & Hayes, 1982). Additional classes
may exist but are of less relevance to this
overview (see Hayes, in press; Zettle &
Hayes, 1982 for further discussion).
Pliance
One type of rule-following has been termed
pliance (Zettle & Hayes, 1982). Pliance is rule-
governed behavior under the control of
socially mediated consequences for a cor-
respondence between the rule and relevant
behavior. The rule itself is termed a ply. For
pliance to be reinforced members of a verbal-
social community must have access to the
relevant ply and be capable of monitoring the
corresponding behavior and controlling rein-
forcing consequences. Reinforcement for
pliance is arbitrary insofar as it is controlled
by socially-mediated reinforcement for a cor-
respondence between the ply and behavior.
For this reason, pliance as a class of rule-
following may occur even when natural (i.e.,
nonarbitrary) contingencies surrounding the
34 ROBERT D. ZETTLE & MARK J. YOUNG
behavior are aversive or punishing. Indeed,
pliance may show the type of insensitivity to
natural contingencies which has been
regarded by some as a defining property of
instructional control (Shimoff, Catania, &
Matthews, 1981).
Following orders or commands may be
viewed as instances of pliance. Correspond-
ence training between saying and doing
(Risley & Hart, 1968; Rogers-Warren & Baer,
1976) similarly may be conceptualized, at
least in part, as a means of establishing a type
of rule-following in which the ply and cor-
responding behavior are emitted by the same
individual. Once pliance is established as a
response class, rule-following may show
sensitivity to a wide array of social variables.
For example, coping self-statements utilized
in cognitive-behavior modification pro-
cedures (Meichenbaum, 1977) have been
shown to control therapeutic responding
when issued in a public context (i.e., thera-
ist has access to the coping instructions
given) but to be ineffective within a private
context (i.e., subjects are led to believe that
the therapist does not have access to the
instructions) (Hayes & Wolf, 1984; Rosenfarb
& Hayes, 1984; Zettle & Hayes, 1983). Similar
effects also have been noted with goal-setting
and self-reinforcement procedures in which
subjects formulate their own rules or instruc-
tions (Hayes, Rosenfarb, Wulfert, Munt,
Korn, & Zettle, 1985).
Such findings would appear to have direct
implications for human operant research.
Changes to schedule-typical responding
which subjects show upon the withdrawal of
instructions suggest that following such rules
may be interpreted as an instance of pliance
(Hayes et al., 1986). Similarly, Hayes et al.
have suggested that differential effects of
shaped versus experimenter-instructed rules
(Catania et al., 1982) may result from the
social context in which shaping occurs. That
is, interactions between experimenter and
subjects necessary to shape particular
guesses may be the same conditions which
enable such verbal behavior to function as
plys. Further research (Matthews et al., 1985)
suggests that shaped guesses which specify
particular ways of responding (e.g., "press
fast") are more likely to exert such control
than those merely tacting the relevant con-
tingencies (e.g., "the button works after a
random number of presses").
The above analysis suggests that social
variables may exert considerable control over
rule-following in human operant research.
Such sources of control conceivably could be
maximized, for example, by having subjects
publicly state their guesses or rules and by
increasing experimenter surveillance of
responding. Alternatively, variables control-
ling pliance might be minimized by having
subjects make guesses and receive instruc-
tions privately (i.e., lead subjects to believe
that the guesses they make and instructions
received cannot be known by the experi-
menter) and by minimizing (at least from the
subjects' perspective) observations of their
responding.
Tracking
A second type of rule-following which
may affect human operant responding is
tracking. Tracking is rule-governed behavior
"under the control of the apparent corre-
spondence between the rule and the way the
world is arranged" (Zettle & Hayes, 1982, p.
81). The rule itself is termed a track. While
reinforcement for pliance is socially-medi-
ated and arbitrary, reinforcement for track-
ing results from natural contingencies sur-
rounding relevant behavior. Unlike pliance,
tracking is not dependent upon the ability of
members of a verbal-social community to
discriminate the presentation of a rule as well
as monitor and reinforce behavior in corre-
spondence with the rule. For this reason,
tracking may occur in a completely private
context as when individuals consult manuals
or written instructions in guiding their
behavior.
Following instructions, directions, and
advice of others (e.g., "The best way to stop
a car on ice is to pump the brakes") may be
viewed as instances of tracking. Like pliance,
tracking may be conceptualized as a func-
tional class of rule-following. With an appro-
priate reinforcement history, tracking may
come under the control of rules of particular
forms (e.g., descriptions of contingencies) as
well as certain rule-givers. It is important to
note that tracking also may occur under the
control of self-generated rules (Skinner, 1969,
chap. 6). Once individuals have learned to
tact their own behavior and the variables of
which it is a function, they may be able to res-
pond more effectively and/or efficiently by
following such rules as tracks. This analysis
 RULE-FOLLOWING AND HUMAN OPERANT RESPONDING
has been cited by Lowe and his colleagues
(Bentall et al., 1985; Lowe et al., 1983) in
accounting for differential response patterns
of preverbal and verbal children on operant
schedules. Only preverbal infants show the
same type of schedule-typical behavior (e.g.,
fixed interval scallops) as displayed by
nonhuman organisms. Older children ex-
hibit response patterns typical of adult sub-
jects as well as verbal behavior which
generally corresponds with their behavior.
The inherent difficulties in relying on post-
session verbal reports as evidence of track-
ing in human operant responding have been
discussed at length by Shimoff (1986) and
only will be reiterated briefly here. Pertain-
ing to the current analysis, it is possible that
verbal behavior reported at postsession may
have functioned as tracks during the session
itself. In the absence of a functional analysis,
however, it is equally plausible that this
verbal behavior may follow rather than
precede and control responding. In doing so,
such verbal behavior represents tacts under
the control of schedule performance.
Another possibility is that verbal behavior
obtained in postsession reports is occasioned
more directly by the experimenter's inquiry.
Because of social contingencies for a cor-
respondence between doing and saying,
subjects' verbal behavior under such condi-
tions could be viewed as impure tacts at least
partially under the control of experimental
demand characteristics.
A METHODOLOGICAL PROPOSAL
Whatever subjects may say to themselves
while responding on operant schedules can
only affect schedule performance by par-
ticipating in a controlling rather than causal
relationship. As detailed by Skinner (1953,
chap. 15) in his analysis of self-control, one
response may enter into a controlling relation-
ship with another, but only when external
contingencies are present which support a
behavior-behavior relationship. In everyday
experience guesses, rules, instructions, and
other types of self-talk control our actions (to
the extent that they do so) because such
verbal control allows us to respond more
effectively and efficiently. When faced with
a problematic situation, some of us are likely
to formulate hypotheses or rules and test
them out. If they prove useful, we are more
35
likely to formulate rules and follow them on
future occasions.
These observations as well as the concep-
tualization of tracking presented suggests a
methodology for studying rule-following in
the human operant laboratory. With the
assistance of a microcomputer it is possible
to match subjects' guesses about operant
schedules with actual programmed con-
tingencies. In such an arrangement, the con-
sequence which reinforces tracking
presumably is "being right." The effect
expected may be similar to that which might
be seen in someone who successfully uses a
formula in playing the horses. An interesting
question is what occurs when the individ-
ual's guesses are no longer correct? Stated
somewhat differently, what will occur when
the natural contingencies which have sup-
ported a controlling relationship between
self-formulated tracks and behavior are
weakened or no longer in effect? Further,
how will an individual's behavior under
such conditions compare with another
whose behavior has not been guided by rules
but has received the same pattern of
reinforcement?
These issues were addressed in a study
described in the next section. The research
was designed to explore the utility of our
analysis of tracking. One group of subjects
undertook a computerized task in which
their guesses about correct responding
initiated programmed contingencies that
matched their guesses. A second group
served in a control condition in which move-
ment of the marker in the task and receipt of
points were yoked with patterns occurring
for experimental subjects. Extinction was
then implemented in both groups. During
the preextinction phase, subjects in the
experimental condition were exposed to con-
tingencies designed to reinforce a control-
ling relationship between their guesses and
related responding. Subjects in the yoked
condition, by contrast, did not have their
tracking reinforced in any systematic
fashion. Based on the analysis of tracking
presented earlier, it was predicted that sub-
jects in the experimental condition, for
whom tracking was reinforced systematically
in the preextinction phase, would show
significantly greater resistance to extinction.
To the extent that results are obtained consis-
tent with this prediction, the validity of our
analysis of rule-following receives prelim-
inary support.
36 ROBERT D. ZETTLE & MARK J. YOUNG
RELATED RESEARCH
Subjects
Sixteen volunteers recruited from an intro-
ductory psychology class served as subjects.
Half were assigned randomly to an experi-
mental condition with the remainder serv-
ing as yoked controls.
Apparatus
An Apple II microcomputer controlled all
schedules and recorded all data. Subjects sat
alone in a small room in front of the com-
puter screen. A shroud over the keyboard
restricted access to all but the space bar and
one key to each of its sides.
Procedure
All subjects initially were presented gen-
eral instructions by the experimenter which
in part read:
The purpose of this study is to understand
further various strategies individuals use
in problem solving. This is not a
psychological test; rather, our interest is in
gaining further insight into how people
learn to solve problems. The task you will
be presented involves learning what is
necessary to move a marker across the
computer screen. The marker may be
moved by pressing the space bar in par-
ticular ways. You will receive one point
each time the marker is moved completely
across the screen from left to right. You
also will receive a message on the screen
indicating the receipt of each point. Each
point will entitle you to one ticket in a
drawing for a $20 prize that will be held
at the end of the study.
After collecting watches, notebooks, and
other distracting materials, the session
began. In order to minimize contamination
by social contingencies all further instruc-
tions were presented on the computer
screen.
The task itself involved movement of a
marker through a row of four boxes
displayed on the computer screen. A point
was awarded for each movement of the
marker out of the extreme right-hand box. A
message then appeared on the screen indi-
cating that a point had been earned and
informing subjects of their cumulative totals.
Movement of the marker was programmed
according to three different schedules. All
subjects initially were presented with a
random time (RT) schedule for 2.5 min.
Movement of the marker occurred after the
passage of a randomly selected interval from
10-40 seconds and was not controlled in any
way by pressing the space bar. After 2.5 min
exposure to RT, the following message
appeared on the screen to subjects in the
experimental condition: "What do you feel
is the controlling factor affecting movement
of the marker?" Subjects were presented
with two options, either "the number of
presses you make" or "the amount of time
between successive presses." Subjects were
instructed to make their choice by pressing
the appropriate key to either side of the space
bar. If subjects indicated "number of
presses," a fixed ratio (FR 18) schedule was
initiated and if "amount of time," a
differential-reinforcement-of-low-rate (DRL
2-s) schedule was implemented for the next
2.5 min. After 2.5 min of either FR or DRL all
subjects received a message on the screen in-
forming them of the termination of that task
and the presentation of a new one. This
message was followed by another 2.5 min of
RT, another request for sujects in the ex-
perimental condition to indicate their
guesses, and depending on their selection,
another 2.5 min of FR or DRL. This sequence
continued for exactly 32 min at which time
extinction was implemented for 15 min. At
the end of the extinction phase all subjects
completed a postsession questionnaire to
provide a manipulation check.
Each subject in the control group was
yoked with another in the experimental con-
dition. During the preextinction phase for
each experimental subject, the computer
recorded the precise movement of the
marker and receipt of points in time. This
information was stored and used in present-
ing the exact temporal sequence of marker
movements and point earnings to yoked sub-
jects as experienced by their experimental
counterparts. During periods in which
experimental subjects were asked to make
guesses yoked subjects received a message
informing them of a short pause before the
next task.
RESULTS
Number of responses for subject pairs dur-
ing preextinction and extinction phases are
presented in Table 1. More preextinction
responses were displayed by subjects in the
yoked control condition for seven of eight
pairs. A randomization test (Siegel, 1956)
 RULE-FOLLOWING AND HUMAN OPERANT RESPONDING 37
Table 1
Number of Responses in Preextinction and Extinction and Response Ratios for Subject Pairs

Condition
Experimental Yoked Control
a~~~~~~
Pair Preextinction Extinction Ratio Prtextinction Extinction Ratio

1 216 115 .532 324 97 .299

2 676 325 .480 1412 684 .484

3 516 292 .566 1457 713 .489

4 584 550 .942 1219 239 .196

5 1353 611 .456 3882 1598 .412

6 928 497 .536 2123 431 .203

7 756 517 .684 275 94 .342
8 441 249 .565 770 220 .286

aValues derived by dividing number of extinction responses by number of preextinction responses.

indicated significantly more responses for
control subjects (p = .02). Because yoked
subjects received the same number of points
as their experimental counterparts, higher
rates of responding consequently were rein-
forced during preextinction for control sub-
jects. Accordingly, higher rates of responding
during extinction would be expected for
these subjects. This, however, was not the
case as higher rates of responding for yoked
subjects were noted in only three of eight
pairs.
The central prediction which this study
sought to evaluate was that subjects in the
experimental condition, because of syste-
matic reinforcement of tracking, would show
greater resistance to extinction. A ratio of
number of extinction responses divided by
number of preextinction responses was
calculated for each subject as an index of
resistance to extinction. These ratios are
displayed in Table 1, with higher values indi-
cating greater resistance to extinction.
Expected results were obtained as higher
ratios were noted for experimental subjects
in seven of eight pairs. A randomization test
also indicated a statistically significant dif-
ference in response ratios between the two
conditions (p = .008).
Postsession reports. At postsession all sub-
jects were asked what they believed was
required to move the marker. Five of the
yoked control subjects indicated that number
or pattern of presses on the space bar was
necessary. The remaining three subjects re-
ported no knowledge of what was required.
No differences were found in comparing
these two subgroups on number of
responses in preextinction, extinction, or in
resistance to extinction ratios.
Half of the experimental subjects indicated
that movement of the marker depended
upon time alone or time between bar
presses. The other half reported number of
responses as well as a temporal requirement.
A comparison of these two subgroups
revealed no differences in number of
responses in preextinction and extinction or
in resistance to extinction ratios. Unfor-
tunately, due to the limited capacity of the
microcomputer it was not possible to deter-
mine whether postsession responses cor-
responded with guesses during the preex-
tinction phase.
SUMMARY AND CONCLUSIONS
The overall findings appear consistent
with our conceptual analysis of tracking as
38 ROBERT D. ZETTLE & MARK J. YOUNG
a functional class of rule-following. Because
reinforcement in the yoked condition was
more intermittent and associated with signif-
icantly higher rates of preextinction respond-
ing, greater resistance to extinction should
have been evidenced by control subjects. The
exact opposite occurred, however, as subjects
in the experimental condition, whose gues-
ses were always correct during preextinction,
demonstrated greater resistance to extinc-
tion. Quite likely yoked subjects also were
making guesses and formulating hypotheses
about schedule requirements. Such guesses
though were not reinforced systematically by
being correct and thus were not expected to
function as tracks.
Apparently the behavior of experimental
subjects was controlled not only by pro-
grammed contingencies but also by nonarbi-
trary reinforcement for responding in corre-
spondence with their guesses. Due to the
exploratory nature of this study and lack of
appropriate control manipulations, other
interpretations may be proposed. It might be
argued, for example, that the process of mak-
ing explicit guesses, regardless of whether or
not they proved to be "correct," accounted
for differences in resistance to extinction. Is
the mere formulation of a rule sufficient or
is a history of correspondence between that
rule and behavioral outcomes also necessary
for a resistance to extinction effect? This issue
could be evaluated by ensuring that guesses
for some subjects are never correct. A more
comprehensive analysis could evaluate the
impact of a wider range of reinforcement
probabilities (e.g., 100%, 75%, 50%, 25%,
0%) on tracking ranging from continuous to
intermittent. In the present study ambiguity
was created by periodic implementation of
an RT schedule. Another strategy to analyze
the degree of control exerted by mere guess-
ing would be to have subjects make repeated
guesses while continuously presented with
an RT schedule.
Consistent with the pliance versus track-
ing distinction, an attempt was made to
minimize social contingencies affecting rule-
following. Because variables controlling
pliance can be fairly subtle, the possibility
that the behavior of experimental subjects
was at least partially affected by social
variables cannot be ruled out. This possi-
bility could be evaluated further by having
some subjects make their guesses under con-
ditions of social surveillance (e.g., have the
experimenter present while subjects make
guesses or have subjects report their guesses
to the experimenter). Any differences noted
between such a manipulation and the experi-
mental condition in the present study would
suggest dissimilar types of rule-following.
Relative to recent differential effects associ-
ated with performance and contingency de-
scriptions (Matthews et al., 1985), it should
be noted that guesses subjects made resemb-
led the latter more than the former. Spe-
cifically subjects were asked to indicate
whether they believed movement of the
marker was dependent on "the number
of presses you make" or "the amount of time
between successive presses." It might be
instructive, therefore, to repeat the pro-
cedure with subjects asked to select among
options which are more descriptive of
performance requirements (e.g., "press
fast" or "press slowly") than the actual
contingencies.
Despite being expected, the findings
reported should be viewed as tentative and
preliminary in nature. In our view, the
results may be regarded more valuably as an
initial demonstration of a methodological
strategy for studying rule-following in
human operant preparations, than as strong
support for our analysis of tracking. The con-
tinued use of microcomputers in the manner
employed in this investigation appears to
have considerable promise in not only fur-
ther clarifying the present findings, but also
in analyzing rule-governed behavior more
generally. Indeed, its applicability may be
limited only by the ingenuity and computer
programming skills of researchers. Whether
our optimism in this regard is warranted or
not, of course, must await further application
of the procedure. However, given the com-
plexity of verbal control and its importance
in predicting and controlling human behav-
ior, any methodological strategy which may
permit a further understanding of rule-fol-
lowing would appear to merit further
exploration.
REFERENCES
Baron, A., & Galizio, M. (1983). Instructional control of
human operant behavior. Psychological Record, 33,
495-520.
Bentall, R. P., Lowe, C. F., & Beasty, A. (1985). The role of
verbal behavior in human learning: H. Developmen-
tal differences. Journal of the Experimental Analysis of
Behavior, 43, 165-181.
 RULE-FOLLOWING AND HUMAN OPERANT RESPONDING
Catania, A. C., Matthews, B. A., & Shimoff, E. (1982).
Instructed versus shaped human verbal behavior:
Interactions with nonverbal responding. Journal of the
Experimental Analysis of Behavior, 38, 233-248.
Galizio, M. (1979). Contingency-shaped and rule-gov-
erned behavior: Instructional control of human loss
avoidance. Journal of the Experimental Analysis of
Behavior, 31, 53-70.
Harzem, P., Lowe, C. F., & Bagshaw, M. (1978). Verbal
control in human operant behavior. Psychological Rec-
ord, 28, 405-423.
Hayes, S. C. (Ed.). (in press). Rule-governed behavior:
Cognition, contingencies, and instructional control. New
York: Plenum Press.
Hayes, S. C., Brownstein, A. J., Zettle, R. D.; Rosenfarb,
I., & Kom, Z. (1986). Rule-governed behavior and sen-
sitivity to changing consequences of responding.
Journal of the Experimental Analysis of Behavior, 45, 237-
256.
Hayes, S. C., Rosenfarb, I., Wulfert, E., Munt, E. D.,
Kom, Z., & Zettle, R. D. (1985). Self-reinforcement
effects: An artifact of social standard setting? Journal
of Applied Behavior Analysis, 18, 201-214.
Hayes, S. C., & Wolf, M. R. (1984). Cues, consequences,
and therapeutic talk: Effects of social context and cop-
ing statements on pain. Behaviour Research and Therapy,
22, 385-392.
Lowe, C. F. (1979). Determinants of human operant
behaviour. In M. D. Zeiler & P. Harzem (Eds.),
Advances in analysis of behavior: Vol. 1. Reinforcement and
the organization of behaviour (pp. 159-192). Chichester,
England: Wiley.
Lowe, C. F. (1983). Radical behaviorism and human
psychology. In G. C. L. Davey (Eds.), Animal models
of human behavior: Conceptual, evolutionary, and
neurological perspectives (pp. 71-93). Chichester, Eng-
land: Wiley.
Lowe, C. F., Beasty, A., & Bentall, R. P. (1983). The role
of verbal behavior in human learning: Infant perfor-
mance on fixed-interval schedules. Journal of the
Experimental Analysis of Behavior, 39, 157-164.
39
Matthews, B. A., Catania, A. C., & Shimoff, E. (1985).
Effects of uninstructed verbal behavior on nonverbal
responding: Contingency descriptions versus perfor-
mance descriptions. Journal of the Experimental Analysis
of Behavior, 43, 155-164.
Meichenbaum, D. (1977). Cognitive-behavior modification:
An integrative approach. New York: Plenum Press.
Risley, T. R., & Hart, B. (1968). Developing corre-
spondence between the non-verbal and verbal
behavior of preschool children. Journal of Applied
Behavior Analysis, 1, 267-281.
Rogers-Warren, A., & Baer, D. M. (1976). Correspond-
ence between saying and doing: Teaching children to
share and praise. Journal ofApplied Behavior Analysis,
9, 335-354.
Rosenfarb, I., & Hayes, S. C. (1984). Social standard
setting: The Achilles heel of informational accounts
of therapeutic change. Behavior Therapy, 15, 515-528.
Shimoff, E. (1986). Postsession verbal reports and the
experimental analysis of behavior. TheAnalysis of Verbal
Behavior, 4, 19-22.
Shimoff, E., Catania, A. C., & Matthews, B. A. (1981).
Uninstructed human responding: Responsivity of
low-rate performance to schedule contingencies. Jour-
nal of the Experimental Analysis of Behavior, 36, 207-220.
Siegel, S. (1956). Nonparametric statistics for the behavioral
sciences. New York: McGraw-Hill.
Skinner. B. F. (1953). Science and human behavior. New
York: Free Press.
Skinner, B. F. (1969). Contingencies of reinforcement: A
theoretical analysis. New York: Appleton-Century
Crofts.
Zettle, R. D., & Hayes, S. C. (1982). Rule-governed
behavior: A potential theoretical framework for
cognitive-behavioral research and therapy. In P. C.
Kendall (Ed.), Advances in cognitive behavioral research
and therapy (Vol. 1, pp. 73-118). New York: Academic
Press.
Zettle, R. D., & Hayes, S. C. (1983). Effects of social
context on the impact of coping self-statements.
Psychological Reports, 52, 391401.